Writing k6 scripts

1. Anatomy of a script #

Init, setup, default, teardown.

First script #

import http from 'k6/http';
import { check, sleep } from 'k6';

export const options = {
  vus: 10,            // utilisateurs virtuels simultanés
  duration: '30s',
};

export default function () {
  const res = http.get('http://localhost:8080/get');
  check(res, {
    'statut 200': (r) => r.status === 200,
    'moins de 300 ms': (r) => r.timings.duration < 300,
  });
  sleep(1);           // temps de réflexion de l'utilisateur
}

k6 run script.js

Good practice: A check does not fail the test: it counts successes and failures. To fail a test, use thresholds.

2. Realistic requests #

Data, headers, groups.

Examples #

import http from 'k6/http';
import { group } from 'k6';

export function setup() {        // exécuté une fois : par exemple, obtenir un jeton
  const res = http.post('http://api/login', JSON.stringify({ user: 'demo', pass: 'demo' }),
                        { headers: { 'Content-Type': 'application/json' } });
  return { token: res.json('token') };
}

export default function (data) {
  const params = { headers: { Authorization: `Bearer ${data.token}` }, tags: { name: 'commande' } };
  group('parcours achat', () => {
    http.get('http://api/produits');
    http.post('http://api/panier', JSON.stringify({ id: 42 }), params);
  });
}

Common mistake: URLs with identifiers (/produits/123) create one metric series per URL. Give them a common name tag ({ tags: { name: 'produit' } }) to keep results readable.

3. Scenarios and executors #

Modeling the load.

VUs or rate? #

Executors decide how load is generated. Two families:

Good to know: With a closed model, if the service slows down, each VU makes fewer requests: the load drops and hides the problem. An open model (arrival rate) keeps the rate and reveals saturation.

Several scenarios #

export const options = {
  scenarios: {
    navigation: {
      executor: 'ramping-vus',
      startVUs: 0,
      stages: [
        { duration: '2m', target: 50 },   // montée
        { duration: '5m', target: 50 },   // plateau
        { duration: '1m', target: 0 },    // descente
      ],
      exec: 'parcours',
    },
    api: {
      executor: 'constant-arrival-rate',
      rate: 200, timeUnit: '1s',          // 200 requêtes par seconde
      duration: '8m',
      preAllocatedVUs: 50, maxVUs: 300,
      exec: 'appelApi',
    },
  },
};
export function parcours() { /* ... */ }
export function appelApi() { /* ... */ }